Optimize (>>) for Seq

[treeowl]

In base 4.8, (*>) will be implemented (by default) using (<$), for which Seq, at least, has an optimized implementation. But this won't help (>>) unless it's defined explicitly.

[foxik]

Patches welcomed :-)

[treeowl]

I'll certainly try. In any case, I realized later that we can do better than using the default *>—in particular, xs *> ys should really be implemented directly, and then >> should be defined in terms of that. There's no reason to calculate id <$ xs at all; instead, we should glue length xs copies of ys together. If there's a magical way to do this, as there is for replicate, then so much the better.

[treeowl]

Fixed by #72